3D Model Generation with Two-Stage Neural Refinement
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current three-dimensional model generation methods using generative adversarial networks (GANs) and autoregressive models produce models with poor quality and low precision.
Innovation Solution
A method involving a two-phase process using a first subnetwork model to generate a rough three-dimensional model and a second subnetwork model to optimize it, with specific generation intensities and seeds, ensuring the final model meets preset resolution conditions and belongs to the desired category.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If a single generative model (GAN or autoregressive model) is used to generate three-dimensional models, then the generation process is simple, but the quality and precision of the generated models are poor
Solution Approach 1:
The patent divides the three-dimensional model generation process into two distinct phases: a first phase using a first generative model to generate an initial three-dimensional model, and a second phase using a second generative model to optimize and refine the model. This segmentation allows each model to specialize in specific aspects of generation, improving overall precision while managing complexity through modular architecture.
Solution Approach 2:
The first three-dimensional model generated by the first generative model serves as an intermediary that bridges the gap between simple generation and high-quality output. This intermediate model provides a foundation that the second generative model can refine, enabling progressive improvement from a basic structure to a high-precision final model.
2Manufacturing precision
If multiple processing rounds are performed to improve model quality, then the generation precision increases, but the time consumption increases
Solution Approach 1:
The patent segments the generation process into two phases with different objectives: the first phase generates a coarse model quickly, and the second phase refines it with higher precision. This time-space segmentation allows the system to achieve high precision without requiring excessive total processing time, as each phase is optimized for its specific purpose.
Solution Approach 2:
The first generative model performs a partial generation task by creating only the basic structure of the three-dimensional model, which requires less time. The second generative model then performs the remaining refinement work. This division of labor between partial actions allows the system to balance speed and precision effectively.
Data Source
AI summary
A three-dimensional model generation method includes: obtaining prompt information describing an object category, and determining a target network model matching the object category indicated by the prompt information, the target network model including a first subnetwork model and a second subnetwork model; obtaining a first generation intensity, a second generation intensity, and a generation seed; performing at least one round of first processing based on the first generation intensity and the generation seed by using the first subnetwork model, to obtain an intermediate three-dimensional model; and performing at least one round of second processing based on the second generation intensity and the intermediate three-dimensional model by using the second subnetwork model, to obtain a three-dimensional object model that satisfies a preset resolution condition and that belongs to the object category.